Skip to content
LED Tutorials
Power14 min read

How to power and cable an LED wall

Size circuits for running load and inrush, inject power on a grid, and daisy-chain data without exceeding port budgets.

LED walls fail more often from power and data than from “bad LEDs.” A cabinet that browns out looks like a dead module. A data chain that is too long looks like flicker. This tutorial covers how to estimate load, build a distribution plan, and loom Ethernet so the map you drew in software still exists in copper.

Electrical work must follow local code and a qualified electrician. The math below is for planning conversations, not for signing a panel schedule.

Start with watts, not with “it fit last time”

From the cabinet datasheet, write down:

  • Max / peak watts (all white at full brightness)
  • Typical / average watts (video content is usually far below full white)
  • Input voltage (100–240 VAC is common; some touring systems are 208 V)
  • Inrush or “start” current if published

For circuit planning, do not use only typical watts. A test pattern, a white lyric background, or a calibration flash can hit peak. A conservative house approach:

Planning watts per cabinet = 70–100% of rated max, unless you have a written brightness cap and you enforce it in the processor.

Example: 80 indoor P2.6 cabinets rated 80 W max each.

  • Peak wall = 6400 W
  • At 120 V, current ≈ 53 A if it were one circuit (it must not be)
  • At 208 V, current ≈ 31 A

Add 20% headroom for PSU inefficiency, long cable, and the fact that nameplates lie in both directions.

Inrush can be 1.5–3× running current for a few milliseconds when many PSUs boot together. Staggered power (sequence strips, multiple breakers not slammed at once) prevents the main from thinking you have a short. If the wall takes a full second to come up after you throw one 30 A breaker, that is a clue.

Break the wall into circuits on the same grid as the cabinets

Draw the cabinet grid. Overlay circuits.

Rules of thumb for 120 V / 20 A US circuits (16 A continuous after derating is the conversation to have with the electrician):

  • Indoor fine-pitch cabinets at ~60–100 W: often 8–16 cabinets per 20 A circuit depending on peak.
  • Outdoor high-nit cabinets at 200–400 W: far fewer — sometimes 3–6 per circuit.

Never daisy-chain power through more cabinets than the manufacturer’s power-in / power-out rating. That rating exists because the connector and the internal bus are not an infinite busbar.

Inject power at multiple points on a tall wall so the last cabinet is not sipping through six jumpers. A brown last row is a classic under-injected chain.

Label every power jumper and every distro outlet with the circuit and the cabinet range: Ckt A — Col 1–2. When a breaker trips, you should know which columns go dark.

Distro, connectors, and grounding

Use a distro that matches the venue: Cam-Lok or similar for touring, twist-locks or house circuits for installs. Consumer power strips are not a wall.

  • One wall, one grounding strategy. Follow the electrician. Do not lift grounds to “kill hum.” LED noise is usually data or dimmer-related, not a reason to unground a metal frame.
  • Separate LED power from audio as a courtesy, but share a proper technical ground. Random isolation transforms create more mysteries than they solve.
  • UPS the processor and the sending card. Cabinets can blink and recover; the processor may take a minute to reboot.
  • Outdoor connectors must be rated and not lying in a puddle. Indoor powerCON on a festival deck is a future trouble call.

Measure voltage at the last cabinet under a white test pattern. If you are more than a few volts below the first cabinet, you are starving the chain.

Data: daisy-chain with a budget

Each sending-card output supports a maximum pixel count at a given frame rate and bit depth. The receiving-card tutorial covers mapping; here you size the copper.

  1. Calculate pixels on the chain: cabinets × pixels per cabinet.
  2. Compare to the sending card’s per-port limit (examples vary: hundreds of thousands of pixels is common; do not assume).
  3. If you are within 80% of the limit, add a port. Heat, long cable, and 10-bit video eat margin.
  4. Keep chains short enough that a single failed jumper only takes a column, not half the wall.

Cable type:

  • Cat5e/Cat6 for short cabinet hops. Use stranded touring cable if the wall moves. Solid install cable if it never moves.
  • Do not exceed the Ethernet distance the manufacturer states (often well under 100 m once you count every jumper).
  • Fiber from FOH or the amp room to the wall. Convert once, then copper along the wall.
  • No random gigabit switches in the LED data path unless the control system is designed for them. Many LED protocols are not generic Ethernet.

Dress data away from dimmer packs and from AC running in parallel for 20 m. A crossing at 90 degrees is better than a 15 m parallel hug.

Loom like you will troubleshoot at 6:55 a.m.

  • Leave a service loop at each injection — not a coil that sits on a connector.
  • Strain-relieve at the cabinet, not on the RJ45 tab.
  • Color-code or number ports to match the processor.
  • Keep power and data on opposite sides of the cabinet when the product allows it.
  • Tie cables so a yanked feeder does not take the lock pin with it.

A tidy loom is not aesthetics. It is how you find the one loose coupler in a 96-cabinet wall.

Sequence of operations

  1. All data mapped and jumpers seated (or at least loomed).
  2. Processor on UPS, project loaded.
  3. Energize cabinets in circuit groups. Watch for a single circuit that refuses to start (short, wet connector, failed PSU).
  4. Confirm sending-card link lights.
  5. Bring up a low-brightness test pattern before full white.
  6. Check last-cabinet voltage and last-cabinet data.

Shut-down is the reverse: take content to black, then cabinets, then processor — or follow house policy if they keep the processor warm.

Common mistakes

Sizing from “typical video” only. The first time someone runs a white lyric look, the breaker opens.

One 30 A twist-lock for an entire outdoor IMAG wall. The math was never done; the wall “worked” at 30% brightness.

Power daisy-chain longer than the spec. The last cabinets flicker only on high-brightness content.

Data and power in one undifferentiated bundle through a motor. The motor eats the RJ45.

Guest HDMI on a 50 ft run as program, with LED data sharing a cheap switch. Three problems wearing one trench coat.

No labels. The next tech will remap ports “to be safe” and create a new problem.

Planning worksheet

Copy this onto the plot:

  • Cabinets: ____ × ____ W max = ____ W peak
  • Voltage: ____ V → ____ A peak
  • Circuits: ____ × ____ A (list cabinet ranges)
  • Sending ports: ____ (list cabinet ranges and pixel counts)
  • Fiber / copper transition at: ____
  • Processor UPS: yes / no
  • Last-cabinet voltage under white: ____ V (fill on site)

When the worksheet is filled, the wall has a power and data design. Until then, it has a pile of jumpers.